← Aerospace Engineering Studio
Concept Explainer · Aerospace

Why GPS Satellites Need Relativity — Not Just Orbital Mechanics

Two separate relativistic effects push a GPS satellite's clock in opposite directions — and they don't cancel. The leftover drift is engineered out before the satellite ever leaves the ground.

GPS positioning is, at its core, a timing problem: a receiver measures how long a radio signal took to arrive from several satellites and converts that into distance using the speed of light. That means the whole system lives or dies on how precisely satellite clocks stay in sync with clocks on the ground. It turns out ordinary orbital mechanics — the part everyone expects — is nowhere near the whole story. Two independent predictions of relativity, working in opposite directions and at very different magnitudes, change the rate of a satellite clock enough to break the system in minutes if left uncorrected.

The Setup

Two effects, two different causes, two different signs

A GPS satellite orbits at roughly 20,200 km altitude, moving at about 14,000 km/h relative to an observer standing on Earth's surface. That velocity brings special relativity into play: any clock moving fast relative to an observer is measured by that observer to tick slower. Separately, the satellite sits much farther from Earth's center than the ground does, in a significantly weaker part of Earth's gravity well. That brings general relativity into play: a clock deeper in a gravity well ticks slower than a clock higher up, so the satellite clock — sitting higher, in weaker gravity — runs faster. One effect slows the satellite clock down. The other speeds it up. They are not the same size, and they do not cancel.

Speed slows it down, gravity speeds it up

Net: faster
Earth's surfaceGround clockreference rateGPS satellite · ~20,200 km · ~14,000 km/hSatellite clockSPEED EFFECT · special relativity−7 μs/day (slower)GRAVITY EFFECT · general relativity+45 μs/day (faster)weaker gravity, higher altitude → clock runs faster
Special relativity (speed)
−7 μs/day
From ~14,000 km/h orbital speed relative to the ground.
General relativity (gravity)
+45 μs/day
From ~20,200 km altitude, in much weaker gravity than the surface.
Net, uncorrected
+38 μs/day
Gravity effect is >6× larger — the satellite clock ends up faster.
The Consequence

38 microseconds a day sounds tiny. It isn't.

GPS turns signal travel time into distance using the speed of light — about 300,000 km per second. At that conversion rate, a timing error doesn't stay small. A clock running 38 microseconds/day fast, left uncorrected, translates into a position error that grows by roughly 10 km per day. And because GPS receivers depend on nanosecond-scale timing precision just to correlate and lock onto satellite signals in the first place, that drift degrades useful accuracy far faster than the daily-average number suggests — GPS positioning would become unusable within minutes of a satellite clock running free, not after a full day.

Position error: uncorrected vs. pre-tuned clock

10km5km0Time since launch, uncorrected (0 → 24 hours)Position errorUncorrected~10 km error after 1 dayWith pre-launch clock correctionnear-zero residual error
Nominal clock frequency
10.23 MHz
The design frequency as it would tick on the ground, uncorrected.
Actual pre-launch tuned frequency
10.22999999543 MHz
Deliberately slowed down so it reads correctly once in orbit.
Why this works

The fix isn't software. It's a slower crystal, built in before launch.

Because the net relativistic offset is predictable and essentially constant for a given orbit — the same +38 μs/day, every satellite, every day — engineers don't need to correct for it in real time. Instead, each GPS satellite's onboard atomic clock is manufactured to tick at 10.22999999543 MHz on the ground — very slightly slower than the nominal 10.23 MHz the system is designed around. Once the satellite reaches orbital velocity and altitude, the combined special- and general-relativistic effects speed that slightly-slow clock back up until it matches the nominal 10.23 MHz rate as seen from the ground. The correction is baked into the hardware rather than patched after the fact, which is only possible because both relativistic effects were calculated in advance with enough confidence to build directly into the satellite design.

Common misconception
"Relativity is a theoretical correction — it doesn't really matter for real engineering."

GPS is one of the most concrete, continuously verified real-world engineering applications of relativity that exists. It isn't a rounding-error refinement layered on top of ordinary orbital mechanics — it is a load-bearing part of the system design. Without boththe special-relativistic (speed) correction and the general-relativistic (gravity) correction built directly into the satellite clock rate, civilian and military GPS navigation would not be "slightly less accurate." It would be non-functional — drifting into uselessness within minutes as accumulated timing error outpaces the nanosecond-level precision the whole positioning scheme depends on. Every working GPS fix is, in a very literal sense, relativity confirmed in orbit, thousands of times a day.

Related Concept Explainers
Why Air Suddenly Stops Behaving Like an Incompressible Fluid
Read it →
Orbital Velocity & Period Calculator
Open tool →

Relativity in GPS — Concept Explainer

Explains why GPS satellite clocks require two separate relativistic corrections — special relativity from orbital speed, and general relativity from weaker gravity at altitude — and why those two effects don't cancel, leaving a net drift that satellite clocks are pre-tuned at manufacture to compensate for.

The Special Relativity Effect

A GPS satellite moves at roughly 14,000 km/h relative to an observer on Earth's surface. Special relativity predicts time dilation for any clock moving relative to an observer: the faster clock ticks slower from the observer's frame. At GPS orbital speed, this effect alone slows the satellite clock by about 7 microseconds per day relative to a ground clock.

The General Relativity Effect

GPS satellites orbit at roughly 20,200 km altitude — far enough from Earth's center to sit in a substantially weaker gravitational field than the surface. General relativity predicts that clocks deeper in a gravity well run slower than clocks higher up in a weaker field. Because the satellite sits in weaker gravity than the ground, this effect speeds the satellite clock up by about 45 microseconds per day — more than six times larger than the special-relativistic slowdown, and opposite in sign.

Why They Don't Cancel, and How the Fix Works

The two effects have different physical causes (relative velocity vs. gravitational potential) and different magnitudes, so they don't offset to zero — they combine to a net drift of about +38 microseconds per day, satellite clock running fast relative to the ground. Left uncorrected, that timing error compounds into a position error growing by roughly 10 km per day, and degrades a receiver's ability to correlate signals long before that, making the system unusable within minutes. GPS satellite clocks are pre-tuned before launch — built to run at 10.22999999543 MHz instead of the nominal 10.23 MHz — specifically so that once relativistic effects speed them up in orbit, they read correctly against ground time.

Frequently asked questions

Which relativistic effect is bigger — speed or gravity?

The general-relativistic (gravity) effect is bigger. It adds about +45 microseconds/day, versus about -7 microseconds/day from the special-relativistic (speed) effect — more than six times larger. That is why the net result is a net speedup, not a slowdown.

Why doesn't the satellite just run an onboard software correction instead of a pre-tuned clock?

Because the net relativistic offset for a given stable orbit is essentially constant, it can be corrected once, permanently, in hardware, by manufacturing the onboard oscillator to run at a slightly offset frequency. This is simpler and more reliable than continuously computing and applying a real-time software correction, though additional smaller relativistic corrections (like orbital eccentricity effects) are still applied in the navigation message.

Does relativity affect the receiver on the ground too?

The dominant effects described here are specific to the satellite's different velocity and altitude relative to the receiver. Ground-based receivers are essentially stationary in the reference frame used for GPS timing, so the correction is applied at the satellite rather than the receiver.

Is this really "proof" that relativity is correct?

It's one of the most practical, continuously running confirmations available. GPS only produces accurate positions because both the special- and general-relativistic predictions were used correctly in the system design — an error in either theory's predicted magnitude would show up immediately as a systematic positioning error.

🎓

Try our Aerospace Engineering Studio

More calculators, simulators, and guides for this discipline.

Related tools & guides

Compressible Flow & Shock Waves — Concept ExplainerOrbital Mechanics & Kepler's LawsOrbital Velocity & Period CalculatorISA Standard Atmosphere Calculator